The AIDA Model: Four Stages for One Message, and Who Really Wrote Them
Learn it in a minute, then the part that matters: what you can honestly claim from it, where the evidence under it runs out, and its disputed origin.
AIDA is a four-part checklist for a single message: attention, interest, desire, action. Learning the four takes a minute.
Knowing what you can honestly claim from them takes longer, and that is the part worth having: what a message that does all four earns you, where the evidence under the model runs out, and who wrote it.
That last one has an answer almost everyone gives and nobody can document: St. Elmo Lewis, in 1898.
What the Four Letters Ask You to Do
The four letters name what one persuasive message has to do in order: stop someone who was not looking for you, hold them, make them want the outcome, and ask for one specific act.
Read it as a completeness check on a page, an email or an advertisement and it earns its place. Read it as a description of how people buy and it will let you down.
Each stage here is written the way the earliest sources wrote it.
Attention
Attention is what the first line has to take, and it is taken from someone who was doing something else.
Arthur Frederick Sheldon put it first, in a 1903 supplement to his salesmanship textbook: "ATTENTION, real attention, is the first great step in the practical art, the real work of selling."
The first A is attention, not awareness, and swapping the words quietly changes what the stage asks of you.
Awareness is a state your reader may already be in. Attention is something your subject line has to win in the moment.
Interest
Interest is the reason to keep reading, and it has to be about the reader instead of about you. Dukesmith wrote the sequence rule in the June 1903 issue of Salesmanship: A Monthly Magazine.
His rule was that "arousing interest" must be thoroughly established before trying to "create desire". A message that jumps from a good headline to a reason to buy has skipped the step that makes anyone care.
Desire
Desire is wanting the outcome, which is different from understanding the offer. Sheldon described that difference in 1906, in a passage Iwamoto reproduces, using his own purchase of a bicycle.
Attention developed into an interest in bicycles, he wrote, and interest changed into desire. The desire became so intense that it turned into a resolve to buy, and he paid out the $75 he could not spare for a wheel he could not afford.
That passage is worth borrowing because it is written from the buyer's side.
Your reader has to be able to picture having the thing, not to admire your description of it.
Action
Action is one specific thing to do, named, with nothing else on the page competing for it. Sheldon's 1910 statement of the four is the plainest wording anyone gave them: "First, attention; Second, interest; Third, desire; Fourth, action."
He also named the failure. Many salesmen, he wrote in 1904, can secure a customer's attention, make him considerably interested and even make him desire the goods, and then "do not succeed in closing the order".
His remedy, written a year later in 1905, was blunt. Make the desire so strong that it leads to decision and action, and get the customer to "do it now".
Using It on One Page, One Email, One Ad
Use AIDA on one message at a time, and use it to check completeness rather than to predict a sale. That is the scale it was written at: Dukesmith and Sheldon were describing one conversation with one customer.
The table below is the whole method, and there is nothing else to it. Take the draft in front of you and answer the middle column.
| Stage | The question to ask of your draft | What a message that skips it does |
|---|---|---|
| Attention | Does the first thing they see stop someone who was not looking for you? | Gets skimmed past by people who would have bought |
| Interest | Is there a reason to keep reading that is about them rather than about you? | Wins the click and loses the reader in the first paragraph |
| Desire | Can they picture having the outcome, not only understand the offer? | Explains itself well and moves nobody |
| Action | Is there one named thing to do, with nothing competing for it? | Ends on interest, and the reader agrees and leaves |
One Email, Walked Through All Four
Say a two-person accounting firm is writing one email to its own list about a new filing deadline. Take the same message against each of the four questions in turn.
Attention. The subject line names the deadline and its date, because that is the one thing in the message a reader who was not thinking about tax will stop for. A subject line about the firm stops nobody.
Interest. The opening line says who the deadline applies to, so a reader can tell in one sentence whether it is about them. That is the difference between a reason to keep reading and a newsletter.
Desire. A sentence about not spending the weekend on it does the third job. What the reader wants is the free Saturday, not the form, and the message has to say the Saturday out loud.
Action. This is where that email usually fails, because it ends with three links: a booking page, a guide, and an invitation to reply.
Three choices is the same as none.
Naming one act, and removing the other two, is the whole of what the fourth stage asks for.
Dukesmith and his co-author put the completeness point sharply in 1905. The four "must follow in logical order, or the sale itself will die in the making if any of these elements are missing".
Your draft, against the four states
Take the message in front of you, one page or one email or one advertisement, and answer the four questions in order. Each one is the question this model asks at that stage.
What the checklist says back
Nothing ticked yet. Answer the four for your own draft, in order, and this reads back where the message stops.
Where the Attribution Went Wrong
The 1898 date credited to Elias St. Elmo Lewis rests on no document at all.
Akinori Iwamoto of Kansai University went looking for one. He published the search in 2023, in the Proceedings of the Conference on Historical Analysis and Research in Marketing.
He read the primary sources, including several that earlier researchers had never found, and set them out in five evidence tables. He did not find the document. He did find where the date came from.
The 1898 Date Has No Document Behind It
Strong wrote the sentence everyone has been repeating. In 1925 he stated that Lewis "formulated the slogan, 'Attract attention, maintain interest, create desire,' in 1898", and that Lewis later added the fourth term, "get action".
Iwamoto's finding about that sentence is the whole story. It carried no citation, and Strong never revised it, restating it unchanged in a second 1925 paper and again in 1938.
From there it spread. Barry accepted it in 1987 and Barry and Howard in 1990, both crediting Lewis and neither adding evidence.
Kotler's Marketing Management has carried a citation since 1972 that could lead a reader to believe Strong was the origin.
Iwamoto's own word for what he found is "curiously": no source, in Strong or anywhere else, indicates the existence of any writings by Lewis in 1898.
What Lewis did publish is on the record and it is not the slogan. In 1899 he asked an advertisement to conform to three rules, "1st, to catch the eye of the reader; 2nd, to inform him; 3rd, to make a customer of him".
The same year he wrote that a proposition must be presented so that it will "attract, interest and convince". In 1903 he asked an advertisement to attract a reader, then interest him, then convince him.
His 1908 book on financial advertising builds its chapters on attracting the attention, the interest, and creating the conviction.
Iwamoto reads that book as the first of Lewis's works to recognize the possibility of the model. Even there the explanation runs no further than attention, interest and either desire or conviction.
It reaches action only in passing, in the line Iwamoto singles out as showing much less intention to formulate the model than Sheldon or Dukesmith had.
Attract, interest, convince.
The four-part slogan is in none of it.
A published reference list for the model says the same thing. Smart Insights sets out "The Original Reference Source" as three Lewis items, dated 1899, 1903 and 1908.
There is no 1898 entry in it.
There is one source that shows Lewis using the slogan, and it points the other way.
It is a report of a talk Lewis gave in Rochester, which Wikipedia dates to January 6, 1910. In it he recalled old advertising men hearing a writer say that all advertising must attract attention, maintain interest, arouse desire, get action.
Iwamoto calls it the only source indicating that Lewis ever uttered the phrase, and adds that it is unknown whether Strong had read it. Taken plainly, it has Lewis quoting a formula other people were already using.
The Names the Record Actually Supports
Dukesmith and Sheldon are the names Iwamoto's evidence points to. Both were writing about salesmanship, and both had the four states in print two decades before Strong credited Lewis with them.
Frank H. Dukesmith edited Salesmanship: A Monthly Magazine and was one of its primary writers, publishing 11 short articles on the model across 1903 and 1904.
In its first issue, dated June 1903, he set out "four distinct conditions to be brought about by the salesman every time a sale is made": securing attention, arousing the interest, creating desire, and closing the sale.
He later settled on "conviction" as his fourth word. Wikipedia's article dates his first published version to 1904, and Iwamoto's table puts the four conditions a year earlier, in that June 1903 issue.
Sheldon founded the Sheldon School of Scientific Salesmanship in 1902, and Iwamoto records, from a 1955 profile of him, that it had published more than 40 pocket-sized textbooks by 1906.
Sheldon wrote in 1912 that the school had reached 10,000 students across the United States and abroad within three years of its founding.
His fourth word was "resolve to buy", including in the 1904 lessons, until 1910, when he wrote the four as attention, interest, desire and action.
His contemporaries noticed. Ralph Starr Butler wrote in 1911 that "Mr. A. F. Sheldon was probably the first to point out the presence of these four steps in every sale".
Neither of the first two studies of the model's formation, both published in 1911 by Butler and by Krebs, mentions Lewis at all.
Iwamoto's conclusion is that Dukesmith and Sheldon, not Lewis, contributed decisively to the formulation of AIDA, and that Sheldon can be regarded as its originator.
He does not close the question, and it matters that he does not. He writes that it is uncertain whether Sheldon or Dukesmith formed the idea by referencing Lewis's research, and that if they did, Lewis is one of the originators.
The acronym itself came later still. Wikipedia dates its first appearance to a 1921 Printers' Ink article by C. P. Russell, who suggested remembering the four words by noticing that their initials spell the opera Aida.
Why the Fourth Word Came From Selling, Not Advertising
The model has two lineages, advertising and salesmanship, and only one of them needed a fourth word. That is Iwamoto's second finding, and it explains a feature of the model you use every week.
The advertising writers of the period stopped at conviction. The three-step formula printed unsigned in Printers' Ink on February 9, 1898 asks an advertisement to attract attention and then to interest and convince.
Lewis ends his chapters in the same place.
Conviction is a state of mind. A copywriter whose advertisement leaves the reader believing has done the job the advertisement was bought to do.
A salesperson has not.
Iwamoto's sentence is that "persuading consumers to take action is a more critical concern for salesmanship and salespeople than for advertising and the writers", and Sheldon, who ran a school for salespeople, is the one who made action the fourth element.
So the fourth stage is the one the model was completed to include. If you are going to take one part of AIDA seriously, that is the argument for it being that one.
Use this chart — embed code and citation
<a href="https://neerajjivnani.com/blog/aida-model/"><img src="https://neerajjivnani.com/infographics/aida-model/attribution-chain.png" alt="Diagram of how the "Lewis, 1898" attribution traveled, read left to right: an empty box for 1898 marked no document, a dashed arrow running back to it from Edward K. Strong Jr.'s uncited 1925 sentence, then that claim passed forward to two parallel carriers, the Barry and the Barry and Howard review articles of 1987 and 1990 and Kotler's Marketing Management from 1972 onwards, and on to a final box marked Today, which reads that the 1898 date is now written as established fact and that every citation trail behind it ends at Strong." width="1200"></a>
<p>Chart: <a href="https://neerajjivnani.com/blog/aida-model/">Neeraj Jivnani</a></p>Neeraj Jivnani, "The AIDA Model: Four Stages for One Message, and Who Really Wrote Them", neerajjivnani.com, https://neerajjivnani.com/blog/aida-model/Free to republish with a link back to this page.
Use this chart — embed code and citation
<a href="https://neerajjivnani.com/blog/aida-model/"><img src="https://neerajjivnani.com/infographics/aida-model/what-lewis-published.png" alt="Side-by-side comparison of the four-part slogan credited to Lewis in 1898, attract attention, maintain interest, create desire and get action, against the four formulas Lewis actually published in 1899, 1903 and 1908, each carrying the word it ends on: make a customer, convince, convince and, for the 1908 chapters, conviction, so that not one of them is the slogan, with a note that Iwamoto reads 1908 as reaching action only in passing." width="1200"></a>
<p>Chart: <a href="https://neerajjivnani.com/blog/aida-model/">Neeraj Jivnani</a></p>Neeraj Jivnani, "The AIDA Model: Four Stages for One Message, and Who Really Wrote Them", neerajjivnani.com, https://neerajjivnani.com/blog/aida-model/Free to republish with a link back to this page.
Use this chart — embed code and citation
<a href="https://neerajjivnani.com/blog/aida-model/"><img src="https://neerajjivnani.com/infographics/aida-model/two-lineages.png" alt="Diagram of the model's two lineages as two stacked tracks. The advertising track runs the unsigned 1898 Printers' Ink formula, Lewis in 1903 and Lewis in 1908, and every stop carries the word conviction, with the 1908 card adding that Lewis reaches action only in passing and builds nothing on it. The salesmanship track runs Dukesmith in 1903 marked closing the sale, Sheldon in 1904 marked resolve to buy, and Sheldon in 1910 marked action, which is highlighted as the stop that gave the model its fourth word." width="1200"></a>
<p>Chart: <a href="https://neerajjivnani.com/blog/aida-model/">Neeraj Jivnani</a></p>Neeraj Jivnani, "The AIDA Model: Four Stages for One Message, and Who Really Wrote Them", neerajjivnani.com, https://neerajjivnani.com/blog/aida-model/Free to republish with a link back to this page.
The Four States, in Messages Written Before the Acronym
The four states were being used to write and to judge real advertising and selling before anyone had a name for them, and those early cases are the clearest worked examples on record. All three below are dated, published, and older than the acronym.
Joseph Addison Richards, a New York advertising agent, wrote an advertisement for his own agency in 1893, and Wikipedia's article quotes it in full.
It offers to teach a client "how to attract attention to what is said in your advertisement; how to hold it until the news is told".
It then offers to inspire confidence in what is being said, to whet the appetite for further information, and to lead to a purchase. Wikipedia's article notes that it contains virtually all the steps without putting them in a hierarchy.
The Contest That Scored Advertisements on the Four
A company then had advertisements scored on the four states, in writing. Between December 1899 and February 1900 the Bissell Carpet Sweeper Company ran a contest for the best written advertisement.
Fred Macey, chairman of another firm and considered an advertising expert at the time, was assigned to examine the submissions. He assessed each one in four numbered respects: the advertisement must receive "Attention", then create "Interest", then create "Desire to Buy", then help "Decision".
That is a written scoring rubric, four stages long, applied by an outside judge two decades before anybody wrote the initials down.
A Sales Script Written as the Four Steps
Dukesmith published the third case in 1904, as a worked example of a specialty sale, and he wrote it as a script.
Study the prospect, make the introduction, get ATTENTION, present the article to arouse INTEREST, develop that with testimonials until the customer has a DESIRE to purchase, then settle price against value and close.
What the three have in common is the unit they operate on. Each one judges a single message meeting a single reader, which is where the model works.
The Evidence Under It Is Thinner Than the Fame
The AIDA model is still relevant as a checklist for one message, and not as a claim about how people buy. That distinction is where its evidence runs out.
Empirical studies have found the model a poor predictor of consumer behavior. Wikipedia's article records that, and names the review that made the general case: Vakratsas and Ambler found little empirical support for hierarchical models as a family.
Found's 2026 guide says it more bluntly. There is "extraordinarily little empirical evidence that proves the AIDA model's effectiveness".
What the Recent Papers Are Made Of
Look at what has been published on the model lately and you can see why the support is thin. Two open-access papers built on it both describe their own method as qualitative.
A 2025 paper in the Journal of Data Analysis and Critical Management, on the model in social media marketing, states that it "uses qualitative method including content analysis of social media campaigns and consumer behavior".
A 2026 paper in the ADI Journal on Recent Innovation, on promotional graphic design, employs "a qualitative descriptive approach, combining literature review and visual content analysis".
Neither reports an effect size, a control group or a conversion rate. That is what "little empirical evidence" looks like from the inside: the model is a lens for describing messages, and not a hypothesis these papers set out to test.
Artificial intelligence has not changed that. It changes how fast you can produce and test the material for each stage, and it leaves the model's own standing exactly where it was.
A generator writes and tests opening lines for attention. A language model drafts the interest paragraph against a named reader, and analytics rank which version of a single named act gets taken.
All of that is worth having. None of it is evidence about the model, and a generated first line still has to take attention from somebody who was not looking for you.
Use this chart — embed code and citation
<a href="https://neerajjivnani.com/blog/aida-model/"><img src="https://neerajjivnani.com/infographics/aida-model/what-the-research-measures.png" alt="Panel comparing what two open-access studies built on the AIDA model, published in 2025 and 2026, say about their own method, each quoted in full, against a third panel listing the three measurements neither of them reports: an effect size, a control group and a conversion rate, closing on the standing criticism that empirical studies have found the model a poor predictor of consumer behavior." width="1200"></a>
<p>Chart: <a href="https://neerajjivnani.com/blog/aida-model/">Neeraj Jivnani</a></p>Neeraj Jivnani, "The AIDA Model: Four Stages for One Message, and Who Really Wrote Them", neerajjivnani.com, https://neerajjivnani.com/blog/aida-model/Free to republish with a link back to this page.
The Models That Came After, and What Each One Changes
Each of the later models includes something AIDA leaves out, or orders the states differently.
They belong with AIDA to a family called the hierarchy of effects models, all of which run cognition, then affect, then behavior, and Wikipedia's article groups them that way.
Four are worth knowing, because between them they cover most of the complaints people make about the original:
- AIDAS adds satisfaction after action, which answers the criticism that the model treats the sale as the end of the relationship.
- AISDALSLove adds search before desire, then like or dislike, share, and love or hate after action, which is an attempt to describe how a purchase happens when the buyer can look things up and tell people.
- Lavidge and Steiner's hierarchy of effects, published in 1961, replaces the four with awareness, knowledge, liking, preference, conviction and purchase, splitting the middle into more states than AIDA allows.
- McGuire's model, from 1978, runs presentation, attention, comprehension, yielding, retention and behavior, and it is the one that puts the message itself at the start.
Every one of these inherits the evidence problem of the original, and adding a stage does not fix it. What the extensions are useful for is naming the thing your own message needs and AIDA has no word for.
The Purchases It Describes Badly
AIDA describes a message well and describes several kinds of purchase badly, and the mismatches are predictable enough to check before you reach for it. Three of them matter in practice, and Wikipedia's article gathers the standard statement of each.
The first is what the longer models were built to fix. With no stage after the sale, anything you sell on a subscription is half described.
The second is the assumption of a straight line. Paid search inverts it, because somebody typing a query already has the interest and meets you afterwards, a case Found's 2026 guide picks out for exactly that reason.
The third is the hierarchy itself. Some research suggests people process a promotional message through two pathways at once, thinking and feeling together, rather than climbing a set of stairs.
Where does that leave you? Reach for AIDA when you are writing one message, to one reader, asking for one act, because that is the case it was built on.
Reach for something else when the purchase involves several people over several months, when the buyer already wants the thing and needs a reason to trust you, or when the value of the sale is in what happens after it.
Keeping the Checklist, Dropping the Claim
AIDA is a good checklist and a weak theory. Keep the first, stop making the second, and it will do real work on every message you write.
It will tell you whether your email is complete. It will not tell you whether the email works, and the people who have studied it have been saying so for decades.
The origin matters for the same reason the evidence does. A claim nobody checks gets repeated until somebody opens the source, and the 1898 date was carried for a century on a sentence that cited nothing.
What the record does support is the more useful story anyway. The model came out of selling as much as advertising, from Dukesmith and Sheldon and the salespeople they wrote for, whose work was not finished until a customer had done something.
That is why it has a fourth stage at all, and the fourth stage is the one most drafts drop.